In [2]:
import requests

url = "https://www.openml.org//data/download/22102255/dataset"
r = requests.get(url, allow_redirects= True)
In [3]:
with open("dataset.text","wb") as f:
    f.write(r.content)
In [4]:
import pandas as pd
In [5]:
data = []

with open("dataset.text","r") as f:
    for line in f.read().split("\n"):
        if line.startswith("@") or line.startswith("%") or line=="":
            continue
        data.append(line)
In [6]:
columns =[]

with open("dataset.text","r") as f:
    for line in f.read().split("\n"):
        if line.startswith("@ATTRIBUTE"):
            columns.append(line.split(" ")[1])
In [7]:
with open("df.csv","w") as f:
    f.write(",".join(columns))
    f.write("\n")
    f.write("\n".join(data))
In [8]:
df = pd.read_csv("df.csv")
df.columns = columns
In [9]:
df['t_win'] = df.round_winner.astype("category").cat.codes
In [10]:
import matplotlib.pyplot as plt
import seaborn as sns

# Filter out non-numeric columns
numeric_columns = [col for col in df.columns if df[col].dtype in ['int64', 'float64']]

# Include the boolean column 'bomb_planted' and the target column 't_win' in the list of numeric columns
numeric_columns.extend(["t_win", "bomb_planted"])

# Convert boolean values to numeric (True -> 1, False -> 0)
df['bomb_planted'] = df['bomb_planted'].astype(int)

# Select only the numeric columns from the DataFrame
numeric_df = df[numeric_columns]

# Calculate correlations
correlations = numeric_df.corr()

# Print absolute correlations sorted in descending order
print(correlations['t_win'].apply(abs).sort_values(ascending=False).iloc[:25])

# Plotting heatmap for visualization
plt.figure(figsize=(18, 22))
sns.heatmap(correlations, cmap='coolwarm', annot=True, fmt=".2f")
plt.title('Correlation Heatmap')
plt.show()

t_win                           1.000000
ct_armor                        0.336382
ct_helmets                      0.308255
t_helmets                       0.297458
ct_defuse_kits                  0.291557
t_armor                         0.290753
ct_grenade_flashbang            0.253868
ct_players_alive                0.216798
ct_grenade_smokegrenade         0.209975
ct_weapon_awp                   0.198626
t_weapon_ak47                   0.194147
ct_health                       0.190662
bomb_planted                    0.187101
ct_weapon_m4a4                  0.178008
ct_grenade_hegrenade            0.168781
ct_grenade_incendiarygrenade    0.168517
ct_weapon_ak47                  0.166855
t_grenade_flashbang             0.166839
ct_weapon_sg553                 0.163963
t_weapon_sg553                  0.163709
ct_weapon_usps                  0.152893
t_weapon_awp                    0.149878
t_players_alive                 0.142518
t_grenade_smokegrenade          0.140348
t_weapon_usps                   0.136694
Name: t_win, dtype: float64
In [11]:
selected_columns = []

for col in columns+["t_win"]:
    try:
        if abs(correlations[col]['t_win']) > 0.15:
            selected_columns.append(col)
    except KeyError:
        pass
    
df_selected = df[selected_columns]
In [12]:
plt.figure(figsize=(18,22))
sns.heatmap(df_selected.corr().sort_values(by="t_win"), annot=True, cmap="YlGnBu")
Out[12]:
<Axes: >
In [13]:
df_selected.hist(figsize=(18,22))
Out[13]:
array([[<Axes: title={'center': 'bomb_planted'}>,
        <Axes: title={'center': 'ct_health'}>,
        <Axes: title={'center': 'ct_armor'}>,
        <Axes: title={'center': 't_armor'}>,
        <Axes: title={'center': 'ct_helmets'}>],
       [<Axes: title={'center': 't_helmets'}>,
        <Axes: title={'center': 'ct_defuse_kits'}>,
        <Axes: title={'center': 'ct_players_alive'}>,
        <Axes: title={'center': 'ct_weapon_ak47'}>,
        <Axes: title={'center': 't_weapon_ak47'}>],
       [<Axes: title={'center': 'ct_weapon_awp'}>,
        <Axes: title={'center': 'ct_weapon_m4a4'}>,
        <Axes: title={'center': 'ct_weapon_sg553'}>,
        <Axes: title={'center': 't_weapon_sg553'}>,
        <Axes: title={'center': 'ct_weapon_usps'}>],
       [<Axes: title={'center': 'ct_grenade_hegrenade'}>,
        <Axes: title={'center': 'ct_grenade_flashbang'}>,
        <Axes: title={'center': 't_grenade_flashbang'}>,
        <Axes: title={'center': 'ct_grenade_smokegrenade'}>,
        <Axes: title={'center': 'ct_grenade_incendiarygrenade'}>],
       [<Axes: title={'center': 't_win'}>, <Axes: >, <Axes: >, <Axes: >,
        <Axes: >]], dtype=object)
In [14]:
df_selected.info()

<class 'pandas.core.frame.DataFrame'>
RangeIndex: 122410 entries, 0 to 122409
Data columns (total 21 columns):
 #   Column                        Non-Null Count   Dtype  
---  ------                        --------------   -----  
 0   bomb_planted                  122410 non-null  int32  
 1   ct_health                     122410 non-null  float64
 2   ct_armor                      122410 non-null  float64
 3   t_armor                       122410 non-null  float64
 4   ct_helmets                    122410 non-null  float64
 5   t_helmets                     122410 non-null  float64
 6   ct_defuse_kits                122410 non-null  float64
 7   ct_players_alive              122410 non-null  float64
 8   ct_weapon_ak47                122410 non-null  float64
 9   t_weapon_ak47                 122410 non-null  float64
 10  ct_weapon_awp                 122410 non-null  float64
 11  ct_weapon_m4a4                122410 non-null  float64
 12  ct_weapon_sg553               122410 non-null  float64
 13  t_weapon_sg553                122410 non-null  float64
 14  ct_weapon_usps                122410 non-null  float64
 15  ct_grenade_hegrenade          122410 non-null  float64
 16  ct_grenade_flashbang          122410 non-null  float64
 17  t_grenade_flashbang           122410 non-null  float64
 18  ct_grenade_smokegrenade       122410 non-null  float64
 19  ct_grenade_incendiarygrenade  122410 non-null  float64
 20  t_win                         122410 non-null  int8   
dtypes: float64(19), int32(1), int8(1)
memory usage: 18.3 MB
In [15]:
from sklearn.model_selection import train_test_split

X, y = df_selected.drop(["t_win"], axis=1), df_selected["t_win"]

X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.2)
In [16]:
len(X_test)
Out[16]:
24482
In [17]:
from sklearn.preprocessing import StandardScaler
from sklearn.neighbors import KNeighborsClassifier

scalar = StandardScaler()

X_train_scaled = scalar.fit_transform(X_train)
X_test_scaled = scalar.transform(X_test)

knn = KNeighborsClassifier()
knn.fit(X_train_scaled, y_train)
Out[17]:
KNeighborsClassifier()
In a Jupyter environment, please rerun this cell to show the HTML representation or trust the notebook.
On GitHub, the HTML representation is unable to render, please try loading this page with nbviewer.org.
KNeighborsClassifier()
In [18]:
knn.score(X_test_scaled, y_test)
Out[18]:
0.7644391798055714
In [19]:
from sklearn.model_selection import RandomizedSearchCV

param_grid = {
    "n_neighbors": list(range(5,17,2)),
    "weights":["uniform","distance"] 
}

knn= KNeighborsClassifier(n_jobs=4)

clf= RandomizedSearchCV(knn, param_grid, n_jobs=4, n_iter=3, verbose=2, cv=3)
clf.fit(X_train_scaled, y_train)

Fitting 3 folds for each of 3 candidates, totalling 9 fits
Out[19]:
RandomizedSearchCV(cv=3, estimator=KNeighborsClassifier(n_jobs=4), n_iter=3,
                   n_jobs=4,
                   param_distributions={'n_neighbors': [5, 7, 9, 11, 13, 15],
                                        'weights': ['uniform', 'distance']},
                   verbose=2)
In a Jupyter environment, please rerun this cell to show the HTML representation or trust the notebook.
On GitHub, the HTML representation is unable to render, please try loading this page with nbviewer.org.
RandomizedSearchCV(cv=3, estimator=KNeighborsClassifier(n_jobs=4), n_iter=3,
                   n_jobs=4,
                   param_distributions={'n_neighbors': [5, 7, 9, 11, 13, 15],
                                        'weights': ['uniform', 'distance']},
                   verbose=2)
KNeighborsClassifier(n_jobs=4)
KNeighborsClassifier(n_jobs=4)
In [20]:
knn=clf.best_estimator_
In [21]:
knn.score(X_test_scaled, y_test)
Out[21]:
0.7798790948451924
In [22]:
from sklearn.ensemble import RandomForestClassifier

forest= RandomForestClassifier(n_jobs=4)
forest.fit(X_train_scaled, y_train)
Out[22]:
RandomForestClassifier(n_jobs=4)
In a Jupyter environment, please rerun this cell to show the HTML representation or trust the notebook.
On GitHub, the HTML representation is unable to render, please try loading this page with nbviewer.org.
RandomForestClassifier(n_jobs=4)
In [23]:
forest.score(X_test_scaled, y_test)
Out[23]:
0.8294257005146638
In [24]:
X_train
Out[24]:
bomb_planted ct_health ct_armor t_armor ct_helmets t_helmets ct_defuse_kits ct_players_alive ct_weapon_ak47 t_weapon_ak47 ct_weapon_awp ct_weapon_m4a4 ct_weapon_sg553 t_weapon_sg553 ct_weapon_usps ct_grenade_hegrenade ct_grenade_flashbang t_grenade_flashbang ct_grenade_smokegrenade ct_grenade_incendiarygrenade
82957 0 500.0 100.0 94.0 0.0 1.0 1.0 5.0 0.0 1.0 1.0 0.0 0.0 0.0 5.0 0.0 0.0 0.0 0.0 0.0
59177 0 500.0 500.0 364.0 1.0 4.0 1.0 5.0 0.0 2.0 1.0 2.0 1.0 1.0 3.0 2.0 4.0 4.0 1.0 1.0
28521 0 500.0 100.0 482.0 1.0 5.0 1.0 5.0 0.0 3.0 1.0 0.0 0.0 1.0 3.0 0.0 0.0 5.0 0.0 1.0
29302 0 500.0 0.0 0.0 0.0 0.0 0.0 5.0 0.0 0.0 0.0 0.0 0.0 0.0 4.0 0.0 0.0 0.0 0.0 0.0
41272 0 500.0 387.0 0.0 2.0 0.0 0.0 5.0 1.0 0.0 0.0 2.0 0.0 0.0 5.0 2.0 1.0 0.0 3.0 0.0
... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ...
252 0 473.0 472.0 100.0 5.0 1.0 5.0 5.0 0.0 0.0 1.0 3.0 0.0 1.0 2.0 2.0 1.0 1.0 2.0 1.0
19267 0 500.0 498.0 300.0 5.0 3.0 4.0 5.0 0.0 2.0 2.0 1.0 1.0 1.0 1.0 0.0 3.0 2.0 1.0 1.0
85186 0 500.0 0.0 0.0 0.0 0.0 0.0 5.0 0.0 0.0 0.0 0.0 0.0 0.0 5.0 0.0 0.0 0.0 0.0 0.0
79774 0 500.0 498.0 400.0 5.0 2.0 3.0 5.0 1.0 0.0 2.0 1.0 1.0 0.0 4.0 2.0 5.0 4.0 4.0 4.0
2646 0 500.0 478.0 500.0 5.0 5.0 4.0 5.0 1.0 1.0 2.0 1.0 1.0 3.0 2.0 3.0 5.0 5.0 5.0 5.0

97928 rows × 20 columns

In [25]:
from tensorflow import keras
from keras.layers import Input, Dense
from keras.models import Sequential

# Initialize a Sequential model
model = Sequential()

# Add an Input layer explicitly
model.add(Input(shape=(20,)))  # Assuming the input features shape is (20,)

# Continue adding the rest of the layers
model.add(Dense(200, activation="relu"))
model.add(Dense(100, activation="relu"))
model.add(Dense(100, activation="relu"))
model.add(Dense(1, activation="sigmoid"))
In [26]:
model.compile(loss="binary_crossentropy",optimizer="adam", metrics=["accuracy"])
In [27]:
early_stopping_cb = keras.callbacks.EarlyStopping(patience=5)

X_train_scaled_train, X_valid, y_train_train, y_valid =  train_test_split(X_train_scaled, y_train, test_size=0.15)

model.fit(X_train_scaled_train, y_train_train, epochs=30, callbacks=[early_stopping_cb], validation_data=(X_valid, y_valid))

Epoch 1/30
2602/2602 ━━━━━━━━━━━━━━━━━━━━ 8s 2ms/step - accuracy: 0.7371 - loss: 0.4910 - val_accuracy: 0.7368 - val_loss: 0.4807
Epoch 2/30
2602/2602 ━━━━━━━━━━━━━━━━━━━━ 7s 3ms/step - accuracy: 0.7474 - loss: 0.4700 - val_accuracy: 0.7389 - val_loss: 0.4774
Epoch 3/30
2602/2602 ━━━━━━━━━━━━━━━━━━━━ 7s 3ms/step - accuracy: 0.7511 - loss: 0.4668 - val_accuracy: 0.7425 - val_loss: 0.4688
Epoch 4/30
2602/2602 ━━━━━━━━━━━━━━━━━━━━ 7s 3ms/step - accuracy: 0.7505 - loss: 0.4634 - val_accuracy: 0.7436 - val_loss: 0.4666
Epoch 5/30
2602/2602 ━━━━━━━━━━━━━━━━━━━━ 5s 2ms/step - accuracy: 0.7545 - loss: 0.4629 - val_accuracy: 0.7467 - val_loss: 0.4677
Epoch 6/30
2602/2602 ━━━━━━━━━━━━━━━━━━━━ 5s 2ms/step - accuracy: 0.7559 - loss: 0.4593 - val_accuracy: 0.7471 - val_loss: 0.4662
Epoch 7/30
2602/2602 ━━━━━━━━━━━━━━━━━━━━ 3s 1ms/step - accuracy: 0.7543 - loss: 0.4572 - val_accuracy: 0.7483 - val_loss: 0.4657
Epoch 8/30
2602/2602 ━━━━━━━━━━━━━━━━━━━━ 3s 1ms/step - accuracy: 0.7593 - loss: 0.4562 - val_accuracy: 0.7502 - val_loss: 0.4665
Epoch 9/30
2602/2602 ━━━━━━━━━━━━━━━━━━━━ 3s 1ms/step - accuracy: 0.7599 - loss: 0.4517 - val_accuracy: 0.7500 - val_loss: 0.4646
Epoch 10/30
2602/2602 ━━━━━━━━━━━━━━━━━━━━ 3s 1ms/step - accuracy: 0.7624 - loss: 0.4483 - val_accuracy: 0.7479 - val_loss: 0.4649
Epoch 11/30
2602/2602 ━━━━━━━━━━━━━━━━━━━━ 3s 1ms/step - accuracy: 0.7653 - loss: 0.4429 - val_accuracy: 0.7488 - val_loss: 0.4604
Epoch 12/30
2602/2602 ━━━━━━━━━━━━━━━━━━━━ 4s 2ms/step - accuracy: 0.7711 - loss: 0.4363 - val_accuracy: 0.7523 - val_loss: 0.4617
Epoch 13/30
2602/2602 ━━━━━━━━━━━━━━━━━━━━ 5s 2ms/step - accuracy: 0.7704 - loss: 0.4378 - val_accuracy: 0.7519 - val_loss: 0.4642
Epoch 14/30
2602/2602 ━━━━━━━━━━━━━━━━━━━━ 5s 2ms/step - accuracy: 0.7713 - loss: 0.4342 - val_accuracy: 0.7517 - val_loss: 0.4618
Epoch 15/30
2602/2602 ━━━━━━━━━━━━━━━━━━━━ 5s 2ms/step - accuracy: 0.7778 - loss: 0.4275 - val_accuracy: 0.7532 - val_loss: 0.4599
Epoch 16/30
2602/2602 ━━━━━━━━━━━━━━━━━━━━ 5s 2ms/step - accuracy: 0.7787 - loss: 0.4235 - val_accuracy: 0.7534 - val_loss: 0.4615
Epoch 17/30
2602/2602 ━━━━━━━━━━━━━━━━━━━━ 3s 1ms/step - accuracy: 0.7821 - loss: 0.4183 - val_accuracy: 0.7487 - val_loss: 0.4669
Epoch 18/30
2602/2602 ━━━━━━━━━━━━━━━━━━━━ 3s 1ms/step - accuracy: 0.7821 - loss: 0.4167 - val_accuracy: 0.7566 - val_loss: 0.4651
Epoch 19/30
2602/2602 ━━━━━━━━━━━━━━━━━━━━ 4s 1ms/step - accuracy: 0.7873 - loss: 0.4092 - val_accuracy: 0.7541 - val_loss: 0.4704
Epoch 20/30
2602/2602 ━━━━━━━━━━━━━━━━━━━━ 5s 2ms/step - accuracy: 0.7918 - loss: 0.4065 - val_accuracy: 0.7551 - val_loss: 0.4669
Out[27]:
<keras.src.callbacks.history.History at 0x2c594040c90>
In [28]:
model.evaluate(X_test_scaled, y_test)

766/766 ━━━━━━━━━━━━━━━━━━━━ 1s 1ms/step - accuracy: 0.7589 - loss: 0.4642
Out[28]:
[0.4672425091266632, 0.7574136257171631]
In [29]:
import numpy as np
from tensorflow import keras

# Assuming 'model' is loaded model

# Generate a random sample scenario with the correct shape
random_scenario = np.random.rand(1, 20)  

# Predict the outcome using the model
prediction = model.predict(random_scenario)
predicted_winner = "Counter-Terrorists" if prediction[0][0] > 0.5 else "Terrorists"

print(f"Predicted winner: {predicted_winner}")

1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 99ms/step
Predicted winner: Counter-Terrorists
In [30]:
model.save('my_model.keras')
In [31]:
import ipywidgets as widgets
from IPython.display import display
import numpy as np
from tensorflow.keras.models import load_model
import random

# path to the saved model
model_path = 'my_model.keras'
model = load_model(model_path)

predict_btn = widgets.Button(description="Predict")
output = widgets.Output()

def on_predict_clicked(b):
    
    
    with output:
        output.clear_output()
        print(f"Making prediction with random map: {random_map_choice}...")
        try:
            prediction = model.predict(input_features)
            predicted_winner = "Counter-Terrorists" if prediction[0][0] > 0.5 else "Terrorists"
            print(f"Predicted winner: {predicted_winner}")
        except Exception as e:
            print(f"An error occurred: {str(e)}")

predict_btn.on_click(on_predict_clicked)
In [32]:
import ipywidgets as widgets
from IPython.display import display
import numpy as np
from tensorflow.keras.models import load_model
import random

# path to the model
model_path = 'my_model.keras' 
model = load_model(model_path)

# Define the widgets for input features
time_left = widgets.FloatText(description='Time Left:', value=75)
ct_score = widgets.IntText(description='CT Score:', value=5)
t_score = widgets.IntText(description='T Score:', value=3)
bomb_planted = widgets.Dropdown(description='Bomb Planted:', options=[('No', 0), ('Yes', 1)], value=0)
ct_health = widgets.IntText(description='CT Health:', value=500)
t_health = widgets.IntText(description='T Health:', value=500)
ct_armor = widgets.IntText(description='CT Armor:', value=400)
t_armor = widgets.IntText(description='T Armor:', value=300)
ct_money = widgets.IntText(description='CT Money:', value=8000)
t_money = widgets.IntText(description='T Money:', value=7500)
ct_helmets = widgets.IntText(description='CT Helmets:', value=5)
t_helmets = widgets.IntText(description='T Helmets:', value=4)
ct_defuse_kits = widgets.IntText(description='CT Defuse Kits:', value=1)
ct_players_alive = widgets.IntText(description='CT Players Alive:', value=5)
t_players_alive = widgets.IntText(description='T Players Alive:', value=4)
ct_weapons = widgets.IntText(description='CT Weapons:', value=4)
t_weapons = widgets.IntText(description='T Weapons:', value=5)
ct_grenades = widgets.IntText(description='CT Grenades:', value=3)
t_grenades = widgets.IntText(description='T Grenades:', value=2)

predict_btn = widgets.Button(description="Predict")
output = widgets.Output()

def on_predict_clicked(b):
    # Randomly select a map from the available options and encode it
    map_options = ['de_dust2', 'de_inferno', 'de_overpass']
    random_map_choice = random.choice(map_options)
    map_choice_encoded = map_options.index(random_map_choice)  # Simple integer encoding for example
    
    input_features = np.array([[time_left.value, ct_score.value, t_score.value, 
                                map_choice_encoded,  # Use the encoded random map choice
                                bomb_planted.value, ct_health.value, t_health.value,
                                ct_armor.value, t_armor.value, ct_money.value, t_money.value,
                                ct_helmets.value, t_helmets.value, ct_defuse_kits.value,
                                ct_players_alive.value, t_players_alive.value, ct_weapons.value,
                                t_weapons.value, ct_grenades.value, t_grenades.value]], dtype=float)
    
    with output:
        output.clear_output()
        print(f"Making prediction with random map: {random_map_choice}...")
        try:
            prediction = model.predict(input_features)
            predicted_winner = "Counter-Terrorists" if prediction[0][0] > 0.5 else "Terrorists"
            print(f"Predicted winner: {predicted_winner}")
        except Exception as e:
            print(f"An error occurred: {str(e)}")

predict_btn.on_click(on_predict_clicked)

# Layout the widgets, excluding map_choice
input_widgets = [time_left, ct_score, t_score, bomb_planted, ct_health, t_health, 
                 ct_armor, t_armor, ct_money, t_money, ct_helmets, t_helmets, ct_defuse_kits, 
                 ct_players_alive, t_players_alive, ct_weapons, t_weapons, ct_grenades, t_grenades, predict_btn]

display(widgets.VBox(input_widgets), output)

VBox(children=(FloatText(value=75.0, description='Time Left:'), IntText(value=5, description='CT Score:'), Int…
ss 
Output()
In [ ]: